[Renaissance-Design]

Anyone use it? If so, who do you use and why?

Anyone's clients asked for it?

This post has been edited by Renaissance-Design: 07 February 2012 - 12:30 PM

[nublue]

Renaissance-Design, on 07 February 2012 - 12:29 PM, said:

Anyone use it? If so, who do you use and why?

Anyone's clients asked for it?

For general monitoring we use Nagios (http://www.nagios.org/) which monitors our network and all the individual servers (plus managed VPS servers). We have this linked to an SMS Gateway so we get notified by text if any of our Nagios alerts are triggered.

We use it because you can setup various different alerts, as in not just up/down alerts. So we can set alerts for when there are too many MySQL connections for example, or for when Apache crashes and so on. The text service is great as it can text the various systems administrators when an alert goes off, and they have to acknowledge the alert in order to stop the texts.

The downside (especially for a client) is that it has a big learning curve. We would normally recommend something like Pingdom for a client (http://www.pingdom.com/). As this is a much nicer interface and gives the basic uptime stats and reports any downtime.

I think the main consideration with this sort of thing with a client is to take into account support. As inevitably any down time will cause the client to send an email requesting an explanation. Which then requires research etc...? And many alerts will be false (as in the site was never down, but the connection to it from pingdom was interrupted for some reason).

I hope that helps.

This post has been edited by nublue: 07 February 2012 - 12:43 PM

[Renaissance-Design]

Cheers for that. It's something I'm thinking of hacking together for a bit of fun and to get to grips with a new framework - I notice Chris Coyier did something similar a couple of years ago.

[nublue]

I see...... Well in that case

This would be our top feature list for a really good monitoring system:

• Something that would output user friendly uptime stats onto a HTML server status page. So as well as monitoring uptime it also publishes it. So companies could use it as marketing collateral as well as a functional tool.
  
  
• Something that notifies in a number of formats. SMS Text, email, maybe an app that alerted.
  
  
• Something with granular monitoring, so it monitors a range of website functions (apache, mysql, slow load times etc..) with different limits set for each. (The cheap monitoring tools just ping, but this is pretty useless really!)
  
  
• Something that sent an email to the hosting ticket system to let everyone know that a problem had happened and what it was so the support guys are aware of it when the emails/calls come through.

There's your brief, I expect a preview on Friday!

This post has been edited by nublue: 07 February 2012 - 01:53 PM

[Renaissance-Design]

Ha! No can do there, Fridays belong to a different app.

I'd definitely be looking at publishing/graphing the stats.

SMS and email are already on the cards. Less keen on the idea of app development because there are 4 major smartphone platforms plus 3 desktop, and the fragmentation is incredible. I'm looking at a responsively-designed web app as the most reasonable option there.

Granular monitoring is a good one - I may pick your brains on implementation there.

Something else I was thinking was I'd need to have more than one monitor, with a bit of logic along the lines of:

If site is unreachable by HEAD request, ping IP (to differentiate between a site down in a shared environment and the whole box/VPS being unreachable). Attempt to connect to other monitors. If other monitors are reachable but customer site/IP isn't, trigger alert. Say 3 different monitors on different IP blocks, with a consensus of 2 monitor alerts to trigger an actual customer alert event?

[Jan de Bruin]

www.downnotifier.com is a nice tool.

This post has been edited by Jan de Bruin: 08 February 2012 - 10:31 PM

[Brandon Line]

I always suggest my customers to use uptime monitoring.

My reasons:

1. I don't trust hosting providers that they report the real SLA level each month.

2. Website failures are not only because of servers (check this infographic) - if anything goes wrong with the website itself (templates, code, database) it is good to be the first to know and wake up your webmaster/admin.

3. Probably the most important - any minute of the outage is a loss to website owner's business. And business has to be protected.

Cheers,

[reckless]

Using http://www.pingdom.com they have free account and paid, features are nice and they offer a API

This post has been edited by reckless: 28 February 2012 - 11:30 PM

[Peartree]

We use Zabbix with 10 second intervals to monitor the core networking - switches, physical servers power, RAID, NIC's, ports etc and then Cacti to monitor bandwidth flowing through the routers and switches - in/out. Any problems and we get notified right away and texts get sent to the on-call engineers.

Chris, take a look at Status2k for a monitoring system for your sites etc, I have used it in the past and its a great piece of software (and very cheap) and does pretty much exactly what you have suggested and more.